CMStatistics 2018: Start Registration
View Submission - CMStatistics
Title: Lagged hierarchical semiparametric models for task-based dynamic functional connectivity (dFC) estimation Authors:  Jaroslaw Harezlak - Indiana University School of Public Health-Bloomington (United States) [presenting]
Zikai Lin - Indiana University School of Medicine (United States)
Maria Kudela - Takeda Pharmaceuticals (United States)
Brandon Oberlin - Indiana University School of Medicine (United States)
Joaquin Goni - Purdue University (United States)
David Kareken - Indiana University School of Medicine (United States)
Mario Dzemidzic - Indiana University School of Medicine (United States)
Abstract: Functional Magnetic Resonance Imaging (fMRI) studies are utilized to assess both brain activation and co-activation among brain regions. Data produced in an MRI scanner consist of hundreds of thousands of time series indicating changes in the blood oxygenation level. We developed a method to estimate the co-activation of hundreds of brain regions at the task-, subject- and population-level at both concurrent time points and at the lagged time intervals. Our method utilizes dynamic functional connectivity approximation, time series bootstrap-based uncertainty evaluation and semiparametric mixed model estimation. We assess our methodology via its application to the study of social and heavy alcohol drinkers reaction to different gustatory cues, including beer, Gatorade and water.